Use of Domain Information to Improve the Performance of an
Evolutionary Algorithm

Ricardo Landa Becerra

The main goal of this thesis work is to explore the capacities of cultural
algorithms to add domain knowledge in evolutionary computation. Within our
objectives is to develop a cultural algorithm for constrained optimization, and
other for multiobjective optimization. With a proper desing of the belief space
we expect to obtain competitive results compared with other state-of-the-art
evolutionary algorithms, but reducing the number of fitness function evaluations
needed. In this paper we focus in the algorithm for constrained optimization,
because the development of the algorithm for multiobjective optimzation is an
early stage.